eval-harness

arm

dedup_pinned_f07efa9_v1

One config on fleet_dedup_v1, scored one way; its repeats averaged per item.

dataset
fleet_dedup_v1
scored against
—
runs
1
headline
key_match
last run
2026-10-30 14:31

Inspect it

01metrics

metricmeanitemsbetter
latency_ms0.174100lower

02cost and speed

cost per item
not recorded
speed per item
p50 0.069 · p95 0.17 ms

Per item, over its items: cost as dollars and tokens (a local model is free but still uses tokens), speed as the median and the slow tail (p95). A goal that names this arm weighs them in its decision.

03runs

runmadeitemscosthow
dedup_pinned_f07efa9_v1_20261030T143130Z2026-10-30 14:31100—measured

04config

config dedup_pinned_f07efa9_v1 · adapter examples/signal-fleet-dedup/adapter.py · credentials are never shown

fleet_commit
f07efa9

05conditions

data.items_sha256
b751fc5f940e7ad27f2470029ce7a87409a6b9b832d2da9eaa961eb9363e6a92
deployment.source
params
fingerprint.version
5
harness.code_sha256
0ab386969208f46c7448153422d5a5ab07833d963869b69fc216a6cd28feb110
harness.version
0.5.0
library.anthropic
2.3.0
library.numpy
2.5.4
library.openai
3.41.0
library.rouge-score
0.1.2
machine.cpu_count
32
machine.platform
Linux/x86_64
machine.python
3.14.7
machine.release
7.2.5-3-omarchy
scorer.id
examples/signal-fleet-dedup/adapter.py
scorer.sha256
8104cf634a23b4134744729d631aaef79571f8e0a076ae5fe691d1f039471ff3